We estimated that, in 2016, 187 million people aged 15–49 years experienced HSV-related GUD, which was equivalent to 5.0% of the world’s population. Altogether GUD was associated with an estimated 8300 million person-days with disease globally. Taking into account parameter uncertainty, the percentage with at least one episode of HSV-related GUD ranged from 3.2% to 7.9% (120–296 million). Established (prevalent) HSV-2 infection caused the majority of GUD compared with both genital HSV-1 infection and recently acquired (incident) HSV-2 infection, meaning GUD trends predominately reflected HSV-2 prevalence. This reflects the natural history of genital HSV infection, with HSV-2 frequently recurring even years after infection, while recurrences of genital HSV-1 infection are much less frequent. We estimated that 178 million people aged 15–49 years experienced GUD due to HSV-2 in 2016. This estimate varied between 143 and 201 million when assumptions around the recurrence rate and duration were varied, and could be as low as 79 million if recurrences stopped after 10 years. Altogether in 2016 there were an estimated 5.0 million first episodes due to HSV-2, 2.4 million first episodes due to genital HSV-1, 959 million HSV-2 recurrences and 4.3 million genital HSV-1 recurrences. Consistent with HSV-2 epidemiology, GUD burden was highest in Africa, was higher in women versus men, and increased with age. HSV-2 infection has been shown to increase susceptibility to HIV,14 and this risk may be even higher in the presence of GUD.17 Therefore, the high GUD burden in Africa and in women is particularly concerning as young women in this region are at high risk of acquiring HIV.85
Strengths and limitations
This study represents the first attempt to estimate the burden of GUD due to HSV globally. Our study has several strengths. We used the most recent available WHO estimates of HSV-2 and genital HSV-1 incidence and prevalence and the best available data on natural history parameters from a detailed review of the literature. By reflecting the complex natural history of genital herpes recurrences and incorporating differences between HSV-1 and HSV-2 infection and by time since infection, we generated a useful paradigm for conceptualising the burden of HSV-related GUD. We were able to generate estimates and demonstrate patterns in GUD burden by HSV type, age, sex and WHO region. Importantly, we also considered the contribution of unrecognised infection to disease burden. By highlighting the number of first episodes and recurrences, our estimates can inform the extent to which clinical care is used currently, and the potential for future HSV-2 interventions to impact on the clinical course of infection. There is a lack of dependable, systematic surveillance of GUD or prevalence studies in most settings, and case reporting may not be reliable. Therefore these estimates are a first step in understanding the total burden of GUD, rather than just the limited number of cases seen in clinical care.
Our estimates have some limitations. First, we did not consider the effect of coinfection with HIV on GUD in our estimates. Our literature review identified some GUD natural history data for PLHIV, from which there was some indication that recurrence frequency and duration can be higher in those who are HIV-positive.73 86–90 By not incorporating the effect of HIV infection on GUD, it is likely that we have underestimated GUD burden in settings with high HIV prevalence. However, to allow the natural history of GUD to vary by HIV status, we would have had to estimate the degree of coinfection, factoring in the epidemiological association between the two infections due to shared risk factors and biological effects of each infection on the other,13 91 92 and considered the effect of CD4 count and antiretroviral therapy status on GUD.90 This would have added in complexity and thus uncertainty, and ultimately we erred on the side of underestimating rather than overestimating GUD burden. Our natural history parameters do not account for antiviral use, which may have already led us to overestimate the number of person-days with GUD. Episodic therapy, which is widely used in many countries, has some effect on symptom duration but no effect on the likelihood of subsequent recurrences.55 Daily suppressive therapy, meanwhile, is effective at reducing symptoms and recurrence rate, although it is not available in most countries.93 Another consideration is that our estimates were done at the WHO regional level: HIV prevalence average for the entire WHO Africa region was 3.9% among 15–49 years in 2018,94 but in South Africa for example this figure was 20.4%.95 The HIV–HSV-2 interaction is critically important and this issue should be studied in depth in future, dedicated analyses.
Second, the GUD estimates build on published WHO estimates of HSV-1 and HSV-2 infection, meaning the issues and assumptions affecting the infection estimates, including data availability, generalisability and quality, are carried forward to the GUD estimates.2 Furthermore, HSV-1 infection estimates were not produced separately by sex for all regions, meaning GUD estimates for HSV-1 may not fully capture differences by sex. However, the infection estimates were informed by systematic reviews to August 2018 and represent the best attempt to quantify HSV-2 and genital HSV-1 prevalence and incidence globally by age and sex. In some countries, such as the USA, an ‘epidemiological transition’ has already occurred whereby rates of oral HSV-1 infection during childhood have declined, and rates of genital HSV-1 infection have increased, due to decreased immunity to HSV-1 on entering adulthood possibly combined with increasing rates of oral sex.3 96 In the most recent prospective evaluation of GUD among women with new HSV infection, in North America, 62% of HSV GUD first episodes were caused by HSV-1.35 Our estimate for the Americas was similar, with 55% of first episodes due to HSV-1. Such trends may be occurring elsewhere in the world.97 Although the potential for genital HSV-1 infection postchildhood is uncertain, our analyses suggest that genital HSV-1 only makes a small contribution to all GUD globally, given the vastly greater number of HSV-2 recurrences.
Similarly, a third limitation concerns the availability, quality and representativeness of natural history data. Studies differed on a number of characteristics, including population group (women or men or both, or men who have sex with men), study location, method of identifying lesions (eg, clinician vs self-report) and length of follow-up. Pooling data from these disparate studies may have introduced bias in our pooled parameter estimates; however, if we excluded more studies, we would have further reduced the availability of data, particularly for settings outside of the USA. The vast majority of the data on first GUD episodes and recurrences, which informed our natural history parameters, came from studies in the USA, which may not reflect the natural history of HSV infection elsewhere. To help mitigate some of these issues, we applied specific inclusion and exclusion criteria for data extraction, pooled data from similar populations as far as possible (ie, those with diagnosed vs unrecognised infection), and standardised data prior to pooling (ie, converting medians to means, and calculating annualised recurrence frequencies).
In addition, the natural history data did not always align perfectly with the possible states for GUD. Our estimate that 4.8% of the world’s population had at least one episode of HSV-2-related GUD globally in 2016 is equivalent to 36% of those with prevalent HSV-2 infection. We would expect this figure not to exceed the percentage with a first episode (since by our definition, only those with a first episode can experience subsequent recurrences). Our pooled estimate of the percentage with a first episode is somewhat below this figure (21.0%). However, not all of the studies contributing data to this estimate were rigorously designed to ensure all those with GUD were identified: one study relied on self-reported symptoms, for example,37 while another only considered the 6 months prior to seroconversion.38 Indeed, a rigorous clinical trial of GUD associated with seroconversion found that 36% had symptoms.36 In addition, natural history studies may enrol people with more severe infection, leading to overestimates of the number and duration of recurrences. In our base case estimates, we used the method that we felt most closely aligned available data with the possible clinical courses for GUD. Our sensitivity analysis showed that the estimates were sensitive to the relative percentages that recurrence rates from clinic-based studies and studies of unrecognised infection are applied to. Therefore, new studies of the natural history among all those with HSV-2 infection would be useful.
Fourth, we assumed that both the percentage who experience one or more recurrences in a given year after the first year and the duration of a recurrence are independent of time since infection (generating one pooled recurrence duration estimate using data with any time since infection), although we did allow the recurrence rate to vary over time. The one available study which examined recurrence frequency and duration over a wide range of time since infection found no change in recurrence frequency but a small decline in recurrence duration, leading to an overall decline in the percentage of days with GUD.64 Our pooled parameter estimate for the mean number of days with recurrent GUD due to HSV-2 was actually slightly increased for longer time since infection (although 95% CI had a large overlap), perhaps because of the disparate studies combined, or perhaps because we did not account for decreasing recurrence duration over time. Therefore we may have overestimated GUD burden, as our sensitivity analysis showed that the burden would be lower if recurrences stopped after more than 10 years since acquisition of HSV-2 infection. However, we also did not consider GUD burden in those aged 50 years or over, potentially leading to an underestimation of burden overall rather than an overestimation. We estimated that individuals with HSV-2 infection experience on average 16 days with GUD annually. Recurrence data from a study excluded from the pooling because it was among all those who are HSV-2 seropositive (including those who were asymptomatic) found a median number of annual recurrences of 2.1,98 which when applied to estimates of the mean recurrence duration for those who are HSV-2 seropositive (data also unused), which ranged from 7 to 10 days,99–101 gives an estimate of 15–21 days with GUD annually, which is similar to our estimate of 16 days.
Finally, we did not consider the modifying effect of previous HSV-1 infection on GUD due to HSV-2. Prior studies have shown that existing HSV-1 infection has no effect on subsequent HSV-2 acquisition,102 but that those with pre-existing HSV-1 infection are more likely to have asymptomatic HSV-2 acquisition compared with those who are HSV-1 seronegative.9 The recurrence rate is similar among symptomatic people with and without HSV-1 infection.42 54 For the purposes of this exercise, we assumed an absence of interactions between HSV-1 and HSV-2, which may have led to a slight overestimation of GUD among those with coinfection. Given that the number of people affected by GUD due to genital HSV-1 is small relative to HSV-2, we also assumed that total GUD burden is simply the sum of GUD burden for HSV-2 and GUD burden for genital HSV-1. We further assumed that among HSV-2 seropositive persons, all GUD was related to HSV-2 infection. Some studies of HSV-2 recurrence rate may have inadvertently captured some GUD due to genital HSV-1. However, given that studies specific to HSV-1 found low genital HSV-1 recurrence rates, and HSV-1 infection only rarely follows HSV-2,9 this was unlikely to have been a significant limitation. We further assumed no contribution to GUD from non-HSV aetiologies in those with genital HSV infection. The inclusion of non-HSV-attributable GUD could potentially have led to an overestimation of GUD due to HSV, but this is unlikely to have been a major issue given the types of studies contributing data, many of which were rigorous clinical trials.